Sterling Anderson doesn’t think humanoid robots should be building cars

General Motors’ chief product officer, Sterling Anderson, used the company’s high‑profile New York technology showcase last week to argue that the future of automobile manufacturing lies in artificial intelligence and software, not in the deployment of humanoid robots on the assembly line. Anderson’s remarks, delivered to a packed auditorium of journalists, investors and industry analysts, sparked a lively debate about the role of robotics in a sector already undergoing rapid transformation through electrification and autonomous‑driving technology.The event, staged at a Manhattan convention centre on October 21, was billed as GM’s “intelligence‑defined vehicle” rollout, a phrase that encapsulates the automaker’s strategy to embed AI‑driven design, centralized computing platforms and virtual engineering into every stage of vehicle creation. While the company reiterated its commitment to hands‑free, eyes‑off autonomous driving and to a new generation of high‑energy‑density batteries, Anderson’s focus was on how these innovations could streamline development cycles and reduce reliance on traditional manufacturing methods.“Humanoid robots may look impressive, but they are not the solution to the challenges we face in building the next wave of vehicles,” Anderson said, gesturing toward a screen that displayed a sleek electric SUV model rendered entirely in digital form..

“What we need are intelligent systems that can design, simulate and optimise parts in a virtual environment, and then translate that data into precise, repeatable processes on the shop floor.” He argued that the current generation of industrial robots—already adept at welding, painting and component placement—already performs the bulk of physical assembly work, and that adding anthropomorphic machines would complicate, rather than simplify, production.Anderson’s stance reflects a broader industry trend. Over the past decade, manufacturers have increasingly turned to collaborative robots, or cobots, which work alongside human operators to handle repetitive tasks. However, the introduction of full‑scale humanoid robots—machines designed to mimic human movement and dexterity—remains limited to research labs and pilot projects..

Experts note that while such robots can perform complex manipulations, they are expensive to produce, require sophisticated perception systems and are still prone to failures in unstructured environments.“From a cost‑benefit perspective, deploying a fleet of humanoid robots in a mass‑production line does not make sense at this stage,” said Dr. Lena Morales, a robotics professor at the Massachusetts Institute of Technology. “The return on investment is outweighed by the reliability of conventional industrial arms, especially when you consider the rigorous safety standards automotive plants must meet.” Morales added that the automotive sector’s biggest bottlenecks now revolve around software integration, supply‑chain constraints for semiconductors and the need for rapid redesigns to accommodate evolving battery technologies.GM’s roadmap, unveiled in a series of detailed presentations, seeks to address those bottlenecks through what Anderson called “virtualised engineering.” The company plans to expand its use of digital twins—exact digital replicas of physical components and systems—to run thousands of simulations in parallel, identifying optimal designs before any metal is cut..

Coupled with a new high‑performance vehicle computer architecture, the approach aims to shorten development timelines from the typical 30‑month cycle to under 18 months for selected models.“The speed at which we can iterate in a virtual space is unprecedented,” Anderson explained, pointing to a live demo where engineers adjusted battery pack geometry on a tablet, instantly seeing the impact on range, weight distribution and thermal management. “When the design is finalised digitally, we can feed that data directly into our manufacturing execution system, ensuring that robots on the line have the exact instructions they need, down to the millimetre.” He emphasized that this integration reduces the need for physical prototypes, cuts material waste and accelerates time‑to‑market.Critics, however, caution that an overreliance on software could expose the industry to new vulnerabilities. Cybersecurity specialists warn that as vehicles become more software‑centric, the attack surface for malicious actors expands..

“A highly connected, AI‑driven vehicle platform is a double‑edged sword,” noted Raj Patel, senior analyst at the International Automotive Security Council. “While it offers efficiency gains, it also demands robust safeguards against hacking, data breaches and unintended algorithmic behaviour.” Patel urged manufacturers to balance innovation with rigorous testing and transparent governance frameworks.In response, Anderson highlighted GM’s investment in a dedicated cybersecurity unit, which he said works closely with the company’s AI teams to embed safety checks into every stage of development. “Security isn’t an afterthought; it’s baked into the architecture from day one,” he asserted..

He also referenced a partnership with a leading cloud provider to host vehicle data in secure, isolated environments, enabling over‑the‑air updates without compromising driver privacy.The debate over humanoid robots versus AI‑driven virtual engineering also touches on workforce implications. Union representatives have expressed concern that increased automation could displace skilled assembly workers, while industry groups argue that new technology creates higher‑value roles in software development, data analysis and system integration. Anderson acknowledged the transition, stating that GM intends to retrain existing staff for these emerging positions..

“Our people are the engine of this transformation,” he said. “We are offering pathways for technicians to become robotics programmers, for line supervisors to become data analysts, and for engineers to specialise in AI model training.”As the event concluded, GM’s press release outlined a target of deploying the intelligence‑defined vehicle platform across three of its core brands by 2028, with an ambition to reduce the average vehicle’s carbon footprint by 30 percent over the next decade. Whether the company’s bet on software and centralized computing will outpace the allure of humanoid robots remains to be seen, but Anderson’s message was clear: the next generation of cars will be built by algorithms and digital twins, not by machines that simply look human..